This patent application claims priority from Canadian patent application No. 2,354,614 filed Jul. 30, 2001.
This invention relates to a passive autofocusing system for use in cameras and camcorders.
Passive autofocus systems are widely used in cameras and camcorders for automatically bringing the subject being photographed or filmed into focus. Such systems are termed xe2x80x9cpassivexe2x80x9d as they respond only to light received from subjects in the field of view; they do not generate any pulses or beams of light to assist in autofocusing.
In such cameras, light from the subject enters the camera through a focusing lens and is directed by suitable mirrors and lenses to an autofocus sensor. The autofocus sensor determines a measure of how well the camera is currently focused as a number (the xe2x80x9cdefocus amountxe2x80x9d). Larger values for the defocus amount indicate that the camera is further out of focus; smaller values indicate that the camera is closer to being in focus. The best focus available with particular equipment is obtained when the defocus amount is a minimum.
The user of an autofocus camera can select a tracking mode in which the camera automatically adjusts the position of the focusing lens based on the defocus amount. In this mode, the defocus amount determines by how much the focusing lens should be moved so as to provide the best focus. The movement of the focusing lens is made by a lens drive unit, typically comprising a motor and gear drive mechanism. After each movement of the focusing lens, the cycle is repeatedxe2x80x94a further measurement of the defocus amount is made and another position adjustment is made to further refine the focusing. This loop is repeated endlessly until the user either releases the shutter mechanism to capture an image or switches out of the tracking mode.
The implementation of the tracking mode varies from camera to camera. Designers skilled in the art can provide a variety of control loops to achieve autofocusing based on the defocus amount. The resulting designs typically make use of parameters such as the masses of the focusing lens and the gears, the amount of friction present, the torque available in the drive motor and the nature of feedback employed.
In cameras having a sufficiently short cycle time for a tracking mode cycle and a sufficiently responsive lens drive unit, the adjustments to the focusing lens position can be made to keep track of a moving subject. This technology relies on maintaining an estimate of the velocity of the subject and predicting the change in focusing lens position necessary to have the subject in focus by the time the motor has had a chance to move.
The autofocus sensor can have more than one focus detector located in the field of view. Each focus detector is capable of determining a defocus amount based on the light incident in that part of the field of view. In some cameras, the user selects the focus detector to use; in other cameras, the selection is made by the camera""s microprocessor according to a stored algorithm.
The focus detectors are made of strips of charge coupled device (xe2x80x9cCCDxe2x80x9d) sensor elements. Typically, each of these strips has 50 CCD sensor elements. The incident light produces a voltage on each CCD sensor element proportional to the intensity and duration of exposure of the incident light. The set of voltages appearing on an entire strip of CCD sensors represents a scan of the intensity of light originating from an approximately linear segment of the field of view (linear within expected design and manufacturing tolerances). With, for example, 50 CCD elements in a sensor strip, the resulting scan produces 50 values representative of the intensity. Any prominent feature perpendicular to the linear segment appears as a sharp transition in intensity along the CCD sensor strip.
Focus detection relies on CCD sensor strips used in pairs. When the focus detector is in use, light from a subject area in the field of view enters the camera through the focusing lens and passes through a secondary imaging lens. The secondary imaging lens directs a portion of the beam along two different light paths so that both sensor strips in a pair are illuminated from the same source. The arrangement of the secondary imaging lens and the CCD sensor strips is made so that both CCD sensor strips in a pair display approximately the same voltage scan, but with one scan shifted out of phase with the other. The defocus amount is calculated as a measure of how much one scan is out of phase with respect to the other, a technique known widely as phase detection.
There are a number of techniques for calculating a defocus amount from the voltages on two CCD sensor strips. For example, Xiong and Shafer describe their research into depth estimation in a technical report (Yalin Xiong and Stephen A. Shafer. Depth from Focusing and Defocusing. Technical Report CMU-RI-TR-93-07, The Robotics Institute, Carnegie Mellon University, 1993) available on the Internet at http://www.ri.cmu.edu/projects/projectxe2x80x94365_text.html. However, the exact details of the calculation of the defocus amount are not relevant to the invention described here.
Each camera design and pair of CCD sensor strips has a known reference value for the minimum defocus amount which brings the subject into best focus. A comparison of the measured defocus amount against the reference value provides the data to control the movement of the motor and gear drive mechanism to move the focusing lens.
The focus detection technique works best if there is a large change or changes in the intensity of light measured along the CCD sensor strips. For example, if both scans from a pair of CCD sensor strips are uniform across the entire length of the detector, it is not possible to determine a defocus amount. On the other hand, if the scans both have the form of a step function, the defocus amount can be calculated with greater accuracy.
In a simple configuration, two straight linear strips are mounted horizontally as a focus detector. The defocus amount can be calculated as long as there is some contrast across the horizonxe2x80x94perhaps a flag pole standing against a clear sky. However, if the subject has only a vertical contrast (for example, a venetian blind) the horizontal linear strips will be unable to detect any contrast and will fail to calculate a defocus amount. In that situation, two vertical linear strips may succeed where a horizontal focus detector would fail. In a more elaborate design, the autofocus sensor uses four linear strips arranged as the arms of a cross. The cross-type autofocus sensor provides defocus amounts for both horizontal and vertical orientations, and a standard algorithm selects the arm of the cross that is best able to focus the scene.
However, this cross-type arrangement may still be inadequate. In some scenes, both the purely horizontal and the purely vertical detectors may provide insufficient contrast, in which event the photographer is forced to tilt the camera 45xc2x0 to allow the detector to find adequate contrast and then rotate the camera back to its shooting position. Unfortunately, the time taken to find a suitable autofocus spot and to reposition the camera may cause a photographer to miss an important picture.
Modern top-of-the-line cameras have autofocus sensors with several cross-shaped focus detectors or combinations of cross-shaped and linear strip focus detectors arranged across the field of view. This further increases the likelihood of finding a satisfactory focus spot. However, there is still the possibility that a photographer may miss a picture if a suitable focus is not quickly obtained.
It is an object of the present invention to provide an improved autofocus sensor for use with a variety of cameras using a passive autofocusing system. In particular, the invention improves the camera""s ability to find sufficient contrast in the field of view. The invention comprises novel arrangements of the CCD sensor strips that make up a focus detector and novel placements of the focus detectors across the field of view.
In accordance with the present invention, a focus detector is made up of one or more pairs of CCD sensor strips, each CCD sensor strip being shaped as an arc of a circle with a typical radius of two millimeters. Both sensor strips in a pair provide a scan of the light intensity from a curved segment of the field of view. Techniques for generating the scans from the sensor strips of a focus detector, processing the scans to calculate a defocus amount and selecting between multiple scans are well known in the art and are not elaborated further here.
Focus detectors with CCD sensor strips arranged in straight lines are limited to detecting a contrast in one direction, along the length of the CCD sensor strip. Curved sensor strips overcome this limitation as they are able to discern contrast in a multitude of directions.
In one embodiment of the invention (xe2x80x9cConfiguration Axe2x80x9d), the autofocus sensor is a single focus detector made up of four strips each of which is shaped as a quarter arc of a circle and is located along one of the four quadrants of a circle whose diameter is approximately four millimeters. The focus detector has a generally circular shape and is located at the centre of the field of view. The focus detector produces four scans which are analyzed by standard means to determine a defocus amount.
(In this description, mention of xe2x80x9ccirclexe2x80x9d, xe2x80x9ccircularxe2x80x9d, and similar geometric terms is to be taken as permitting normal design and manufacturing tolerances and approximations, with the recognition that, for example, a circle or portion thereof may be approximated by one or more other geometrical elements.)
In a further embodiment of the invention (xe2x80x9cConfiguration Bxe2x80x9d), the autofocus sensor comprises eight strips each of which is shaped as a quarter arc of a circle. The eight strips are arranged as three focus detectors, as follows:
(1) four strips as in Configuration A;
(2) two strips as a semicircle placed to the left of the central circle; and
(3) two strips as a semicircle placed to the right of the central circle.
In a further embodiment of the invention (xe2x80x9cConfiguration Cxe2x80x9d), the autofocus sensor comprises twelve strips each of which is shaped as a quarter arc of a circle. The twelve strips are arranged as three circular focus detectors, each of which has four strips arranged on the circumference of a circle, located at approximately equal intervals across the horizontal axis of the field of view (i.e., equal intervals within usual design and manufacturing tolerances).
In a further embodiment of the invention (xe2x80x9cConfiguration Dxe2x80x9d), the autofocus sensor comprises sixteen strips each of which is shaped as a quarter arc of a circle. The sixteen strips are arranged as five focus detectors, as follows:
(1) twelve strips as in Configuration C;
(2) two strips as a semicircle placed above the central circle; and
(3) two strips as a semicircle placed below the central circle.
In a further embodiment of the invention (xe2x80x9cConfiguration Exe2x80x9d), the autofocus sensor comprises twenty strips each of which is shaped as a quarter arc of a circle. The twenty strips are arranged as seven focus detectors, as follows:
(1) twelve strips as in Configuration C;
(2) two strips as a semicircle placed above and midway between the central and the left circles;
(3) two strips as a semicircle placed above and midway between the central and the right circles;
(4) two strips as a semicircle placed below and midway between the central and the left circles; and
(5) two strips as a semicircle placed below and midway between the central and the right circles;
In those embodiments where there is more than one focus detector, the selection of focus detector to use is made either by the photographer manually or by a microprocessor using a known algorithm. The calculation of the defocus amount is made by standard means from the scans provided at each focus detector.
The fabrication of curved CCD sensor strips does not present a technical difficulty. Such curved CCD sensor strips can be readily manufactured with existing techniques known to those skilled in the art. Note that curvature of a series of elements can be approximated by sequential slight misalignment of successive elements so that the entire sequence of elements approximates the intended curve. The invention embraces design and manufacturing approximations of this sort.
Configuration E is the preferred embodiment. Although somewhat more complex to manufacture, this configuration provides the best ability to focus on subjects of varied shapes located at different points in the field of view.